A new wavelet estimator for image denoising

摘要

Nonlinear wavelet/wavelet packet (W/WP) estimators are becoming popular methods for image denoising. Most of the related works are based on a hard or soft thresholding of the W/WP coefficients. We present a new nonlinear estimation method whose characteristics are adapted to the statistics of the image under study. We describe the statistical problem which is addressed in this paper and present different methods for adaptive image denoising. We emphasize the shortcomings of classical threshold estimators and propose an alternative approach. We then illustrate the interest of our method by simulation examples on satellite images.